The link between salary and performance: Are NBA players

TALLINN UNIVERSITY OF TECHNOLOGY
 School of Business and Governance
 Department of Finance and Economics

 Daniel Hentilä
 The link between salary and performance: Are NBA players
 overpaid?
 Bachelor’s thesis
Programme International Business Administration, specialization Finance and Accounting

 Supervisor: Pavlo Illiashenko

 Tallinn 2019

I hereby declare that I have compiled the paper independently
and all works, important standpoints and data by other authors
has been properly referenced and the same paper
has not been previously presented for grading.
The document length is 8705 words from the introduction to the end of conclusion.

Daniel Hentilä ……………………………
 (signature, date)
Student code: 166324TVTB
Student e-mail address: hentila.daniel@gmail.com

Supervisor: Pavlo Illiashenko
The paper conforms to requirements in force

……………………………………………
(signature, date)

Chairman of the Defence Committee:
Permitted to the defence
…………………………………
(name, signature, date)

TABLE OF CONTENTS

ABSTRACT .................................................................................................................................... 4
INTRODUCTION ........................................................................................................................... 5
1. BACKGROUND ......................................................................................................................... 7
 1.1. Previous findings .................................................................................................................. 7
 1.2. NBA Collective Bargaining Agreement ............................................................................ 10
 1.2.1. NBA salary cap ........................................................................................................... 11
 1.2.2. NBA salary cap exceptions ......................................................................................... 12
 1.2.3. Luxury tax ................................................................................................................... 13
 1.3. Superstar effect ................................................................................................................... 13
 1.4. Player efficiency ................................................................................................................. 14
2. METHODOLOGY .................................................................................................................... 16
3. EMPIRICAL PART .................................................................................................................. 20
 3.1. Regular season regression .................................................................................................. 20
 3.2. Playoff regression ............................................................................................................... 22
 3.3. Suggested salary ................................................................................................................. 23
 3.4. Limitations ......................................................................................................................... 24
CONCLUSION ............................................................................................................................. 26
LIST OF REFERENCES .............................................................................................................. 28
APPENDICES ............................................................................................................................... 31
 Appendix 1. Definitions ............................................................................................................ 31

 3

ABSTRACT

The National Basketball Association (NBA) is one of the four major sports leagues in North
America. Because of NBA’s growing popularity and increased revenue that it brings, the value of
players’ contracts has increased substantially over the recent decade. This, in turn, started a
discussion if high players’ wages are justified by the players’ performance. With the goal of
contributing to this discussion, the present research explores the relationship between NBA
players’ salary and their on-court performance. The objective of the paper is to find out if salary
has a positive relationship to player performance, measured by Player Impact Estimate (PIE)
rating. The secondary objective is to discover if players in the NBA are paid according to their
performance. Based on the analysis of financial and performance data for the 2017-2018 season
(N=361) the thesis finds that there is a significant positive relationship between player salary and
performance controlling for player’s age, playing position, and team fixed effects. The results also
show that players in the NBA are on average underpaid in relation to their performance.

Keywords: NBA, Salary, Performance

 4

INTRODUCTION

The National Basketball Association (NBA) is one of the four major sports leagues in North
America. NBA has gained a lot of popularity because of superstar player’s as well as international
player’s entering the league. With the growing popularity the revenue that NBA franchises collect
during a season has also been increasing. This has led to a situation where player’s in the NBA are
able to make bigger contracts since teams have more money to spend on players. NBA currently
has the highest salaries of all four major sports leagues. Each team usually has one superstar player
who can earn up to 20%-25% of a team’s total payroll. This is possible due to the fact that NBA
teams only have around 15 players’ instead of over 20 player’s that other leagues such as MLB,
NHL and NFL have. The question that arises from these large contracts is, do player’s in the NBA
perform according to their given salary or do teams hand out large contracts just to acquire the
player’s they want. Should the payroll be distributed more evenly among teams’ players and not
so, that one or two players earn almost half of the whole team’s payroll and the rest of the players
get what is left. Or do the players with high salary actually perform according to their salary and
the rest with smaller salary according to theirs. This research explores the relationship between
player salary and player performance. The question that this research tries to answer is: are player’s
in the NBA under- or overpaid in relation to their on-court performance?

Previous studies that have been made, have tried to identify if players are under- or overpaid
according to their ability to add wins for a team. Tyler Stanek attempted to find if players are
under- or overpaid based on the value of players contribution to team revenue in terms of wins
added (Stanek 2016, 1). Stanek used his own model as well as Arturo Galletti’s model who is a
director of analytics for The Wages of Wins Journal (Stanek 2016, 21). Stanek also used
Hollinger’s Estimated Wins Added (EWA) metric and David Berri’s Wins Produced (WP) metric
to calculate the contribution on team revenue a player has brought. The results Stanek got using
his own model were that, player’s in the NBA are overpaid by approx. $2,5 million according to
EWA approach and $2,2 million according to WP approach. When Stanek applied Galletti’s model
he got the results that with the EWA approach players are $271,764 underpaid and with WP
approach players are $671,691 underpaid (Stanek 2016, 34). The results that Stanek got from his
study are quite different, so it is hard to say what is right. The initial thought is that players who

 5

have higher salaries would also perform better. However, as previous studies have shown, there is
a possible overcompensation of players in relation to their ability to add wins for a team.

Since the results in the existing literature are ambiguous, the aim of this research is to clarify the
link between player salary and individual performance taking into consideration individual player
performance and salary. The players’ performance is measured with Player Impact Estimate (PIE)
index (NBA Stats 2019) while players salary is measured either in millions of U.S. dollars or as a
percentage of player’s salary contribution to the team payroll. Due to the issue of data availability
the sample covers 361 players during only one season, the 2017-2018.

The results received from the regression analysis that examined the relationship between player
performance and salary showed a significant positive relationship. The relationship was measured
on regular season performance as well as playoff performance level and both indicated a positive
relationship. The second objective of this paper was to discover if players in the NBA are paid
according to their performance. This was achieved by firstly calculating teams total PIE score
which consist of adding together PIE scores from players who play in the same team. After this
each player’s individual PIE score was divided by the team PIE in order to obtain players
individual contribution to their team. Lastly, the percentage of player’s salary contribution to the
team payroll was compared to the PIE contribution each player gives to their team PIE. The results
showed that on average players in the NBA are underpaid in relation to their performance. The
most underpaid were younger players in the league while older players were more overpaid.

The paper begins with background section which includes previous findings as well as important
information that is needed in order to understand how the paper. The second part is the
methodology part. Last part is the empirical part which goes over the results, and conclusion which
completes the paper.

 6

1. BACKGROUND

National Basketball Association or NBA is one of the four major sports leagues in North America
in addition to NFL, MLB and NHL. Basketball is becoming the most popular sport in North
America and NBA is attracting players from all over the world. This creates a huge fan base also
outside of North America for NBA since each team has international players. With the rising
popularity of NBA, the teams that play in the NBA are gathering huge revenues each year. This
has allowed NBA to have the largest contracts in all of the four major sports leagues. Teams
allocate their allowed salary cap to players unevenly and some players take around 20 to 25% of
the whole team’s payroll. The contracts are supposedly made with the fact that better performance
equals a larger contract. This first chapter goes through previous studies made on the topic as well
as explains factors that are essential in order to understand how teams pay for their players and
how much are they allowed to pay. Also, is there some other value in a player for a team besides
their performance on the basketball court.

1.1. Previous findings

Previous studies have examined how the salary in the NBA is distributed amongst the players and
what are the traits in players that count when teams offer contracts. The research that is the closest
match to this paper is Tyler Stanek’s “Player Performance and Team Revenues: NBA Player Salary
Analysis”. Stanek analyzes how much revenue can a win bring to a team, and using player marginal
revenue product (MRP), Stanek examines the possible under- or overpayment of players and
superstars in the league (Stanek 2016, 1). Stanek attempts to find if players are under- or
overcompensated for their on-court performance based on the value of player’s contribution to
team revenue in terms of wins added. This approach is inspired by the work of Neale (1964) who
studies the revenue sharing practices in the Women’s National Basketball Association (WNBA),
where players are not employed and do not earn salary from their team but are employed and
compensated by the league. This leads to a situation where players have the ability to earn their
salary as a percentage of the revenue that they have brought to the league. This situation would
end the revenue issue that arises from superstars (Stanek 2016, 21).

 7

Stanek studies the possibility of under- or overpayment of player, relative to their marginal revenue
product (MRP) by the revenue implications that a win has on a team. After finding out the MRP
for a player, Stanek is using John Hollinger’s Estimated Wins Added (EWA) statistics as well as
David Berri’s Wins Produced (WP) metric. These two measurements are used separately in the
study with both multiplied with the sum of Wins and LagWins coefficients. Since lagged wins
have happened in the past, Stanek discounts the lagged wins with a discount rate of 5% to account
for the time value of money. When the MRP has been calculated for both Hollinger’s EWA and
Berri’s WP, Stanek subtracts each player’s MRP from their actual salary in order to find out if
players are under- or overpaid relative to their MRP (Stanek 2016, 26). Stanek calculations for
player MRP are done in the following way.

 = / × ( . . $) (1)

 = / × ( . . $) (2)

The data Stanek used for this study is compiled from seasons between 2009-10 and 2014-15,
excluding season 2011-2012 since NBA had a lockout during that season missing 16 games from
the regular season. For Hollinger’s EWA, the data consist of 231 observations from which 30 are
the top paid player’s and 30 are the top EWA performers. For Berri’s WP, the data consist of 254
observations from which 30 are the top paid player’s and 30 are the top WP performers (Stanek
2016, 31).

Stanek’s results for his question, if player’s in the NBA are under- or overpaid based on Hollinger’s
EWA metric show that on average player’s in the NBA are $2,546,746 overpaid compared to their
actual salary (Stanek 2016, 33-34). In order to find out if player’s in the NBA are under- or
overpaid Stanek used his own calculation of players EWA multiplied with the value of a win from
which player’s actual salary would be subtracted. In a different calculation Stanek used Galletti’s
value of win instead of his own and got the results that players in the NBA are on average $271,764
underpaid relative to their marginal revenue product (Stanek 2016, 33-34).

The results Stanek got from calculations with Berri’s WP metric were that on average player’s in
the NBA are $2,221,592 overpaid relative to their MRP (Stanek 2016, 34). The results between
Hollinger’s EWA and Berri’s WP are more or less similar. Stanek also applied Galetti’s value to
calculate Berri’s WP and got the result that on average NBA players are underpaid by $671,691.

 8

The difference is more drastic between Hollinger’s EWA and Berri’s WP when Galetti’s value of
win is applied (Stanek 2016, 34-36).

After discovering that, based on Players MRP vs. Salary (Hollinger’s EWA approach and Berri’s
WP approach) player’s in the NBA are overpaid, Stanek studies if age has a positive relationship
with salary (Stanek 2016, 36). The results were that both Hollinger’s EWA and Berri’s WP suggests
that as player’s get older in the NBA, they are increasingly overpaid relative to their MRP. These
results support previous studies done by Leeds and Kowalewski (1999) who determined that
veteran players are overpaid while younger players earn less (Stanek 2016, 36). These results can
be somehow explained with the fact that the minimum salary for player’s in the NBA is determined
by the years of experience in the league. Meaning that a player with 5 years of experience will
have a lower minimum salary than a player with 12 years of experience in the league. This still
does not justify the fact that NBA veterans are overpaid relative to their MRP since the differences
in the minimum salary are relatively small.

Other studies that have been made considering NBA player’s performance and their respective
salary are different determinants of salary. Three previous studies that have been made on the
subject of how NBA player’s salary is determined will be used in this paper. Two of these studies
have been conducted by Kevin Sigler with the help of William Compton and William Sackley. The
third study has been done by Tyler Stanek.

Sigler and Sackley in their article, NBA Players: are they paid for performance? study the
relationship between NBA player’s salary and their performance on the court. Sigler and Sackley
studied the relationship with a simple regression equation. The regression equation was the
following: Player Salary = function (REB, ASST, PTS), where the independent variables were
respectively rebounds per game, assists per game and points per game (Sackley, Sigler 2000, 47-
48). The results showed that rebounds and points had a positive relationship at a 95% confidence
level to player’s salary while assists were insignificant. The second study done by Sigler with
William Compton, NBA Players’ Pay and Performance: What Counts? continues Sigler’s prior
studies that showed that points scored, and rebounds had a positive relationship to salary. The
reason for this study is the fact that the NBA is evolving and the determinants for salary are
changing (Compton, Sigler 2018). The multiple regression that was conducted in Sigler’s and
Compton’s study uses NBA players pay from the season 2017-18 as the dependent variable. This
time their independent variables consist of points, experience, field goal %, 3-pointers made,
rebound, assists, blocks, personal fouls and John Hollinger’s Player Efficiency Rating (PER). The

 9

results showed four variables that were significant at the level of 0.01**, these variables were
points, experience, rebounds and personal fouls. Assists per game were significant at the level of
0.05*. Other variables were found insignificant. Sigler and Compton did a backward stepwise
regression for their multiple regression results. They removed one statistically insignificant
variable at a time. This eliminated the variable that had the highest P-value. After the backward
stepwise regression, the same variables points’, experience, rebounds, assists and personal fouls
stayed in the multiple regression results (Compton, Sigler 2018).

1.2. NBA Collective Bargaining Agreement

Collective bargaining agreement (CBA) was created to solve disputes between the national
basketball association and its owners and the players who play in different franchises. The current
CBA agreed between NBA and National Basketball Players Association (NBPA) took effect on
July 1, 2017 and it will be effective for seven years through seasons 2017-18 and 2023-24. Both
parties, NBA and the NBPA have the ability to opt out of the agreement after the season 2022-23
with the condition that they inform the other party no later than December 15, 2022 (NBA Official
2017). The purpose of the agreement is outline rules for player contracts including 10-Day
contracts, two-way contracts etc., minimum and maximum salaries, bonuses, travel expenses, legal
investigations and many other (Caldwell 2010).

To keep this paper simple, this chapter will only go through the rules that are necessary in this
research. The parts from CBA that are necessary for this paper are the NBA salary cap, the
exceptions on the salary cap and luxury tax. Some changes that were made to the new 2017 CBA
were the ability to give six-year extensions to two veteran players. This gives teams a better
opportunity to keep their best players in their roster. The new 2017 CBA also made a rule that will
shorten the preseason and make the regular season to start earlier. This rule was conducted in order
to reduce the travel that teams face during the regular season. In the new 2017 CBA the over-36-
rule was changed to over-38-rule. This rule states that teams are not allowed to offer five-year
maximum contracts to players who are going to turn 38 years old in any part of the contract unlike
the old rule which stated that five-year maximum contracts could not be offered to players turning
36 years old in any point of the contract (NBA Official 2017).

 10

1.2.1. NBA salary cap

The salary cap in the NBA is assigned by the collective bargaining agreement. Salary caps purpose
in the league is to increase parity. To keep any league interesting there has to be competition. Salary
caps allows the league to have a competitive balance and make sure that not every game is
predictable. Salary caps work so that each team is limited to a certain amount of many that they
can pay to their players. This creates a situation where each team has an equal position when it
comes to offering contracts to players. NBA teams are located all over North America in different
cities that are different in sizes. Without the salary cap, big city teams such as Los Angeles Lakers
who come from Los Angeles, California with a population of approximately four million, would
have a much larger local market than Milwaukee Bucks who come from Milwaukee, Wisconsin
with a population of almost 600,000 (Shorin 2017).

The salary cap in the NBA is what is called a soft cap. Soft cap allows teams to exceed their salary
cap with different exceptions. Most of the major leagues in North America have hard caps which
mean that there are very few exceptions on how to exceed the salary cap. NBA’s salary cap is
calculated with ‘Basketball Related Income’ or BRI. BRI means the aggregate operating revenues
for salary cap year generated or to be generated by the NBA, NBA Properties, Inc. and any of its
subsidiaries whether now in existence or to be created in the future, NBA Media Ventures LLC,
and any other entities which are controlled or 50% owned by any of the recently mentioned
companies. This revenue is compiled of regular season gate receipts, net of applicable taxes,
surcharges, imposts, facility fees as well as many other that sources that come from NBA
franchises (Collective Bargaining Agreement 2017, 125-127). The salary cap for each team is
calculated as follows, the salary cap is 44,74% of the projected BRI for the salary cap year, less
projected benefits for such salary cap year, divided by the number of teams scheduled to play in
the upcoming season (Collective Bargaining Agreement 2017, 159-160).

NBA’s salary cap has been rising each year and after 2015-16 season the salary cap has increased
by almost 45%. In 2017-18 NBA’s salary cap was just over $99 million. This means that each team
in the NBA had $99 million to distribute in salaries for their 15-16 players. With the salary cap
there also comes a minimum team salary that each team has to pay for their players. This is again
to keep the competitive balance alive in the league. The minimum salary that each team has to pay
for their players is 90% of the salary cap. If a team fails to meet this minimum salary requirement,

 11

they are obligated to distribute the amount of shortfall to their players who were on the team roster
during the season. The CBA also states that teams shall not exceed the salary cap that is mandated
by the league but there are many different exceptions on how to exceed the salary cap. The salary
cap can be exceeded with these exceptions: Mid-level exception, Bi-annual exception, rookie
exception, two-way contract, disabled player exception, veteran free agent exception, minimum
player salary exception, traded player exception as well as others. (Caldwell 2010).

1.2.2. NBA salary cap exceptions

NBA has many different exceptions how teams are able to exceed the salary cap that is set for each
team. The mid-level exceptions allow teams to sign one or more player contracts during each salary
cap year, not to exceed four seasons in length, that, in the aggregate, provide for salaries and
unlikely bonuses in the first salary cap year totaling up to the amounts set forth. This amount was
$8.406 Million for the 2017-18 salary cap year (Collective Bargaining Agreement 2017, 203-204).
The Bi-annual exceptions allows teams to sign one or more player contracts not to exceed two
seasons in length during each salary cap year. The salaries and unlikely bonuses in the first salary
cap year are allowed to total up to amounts set forth. For the 2017-18 salary cap the Bi-annual
exception amount totaled up to $3,290 million (Collective Bargaining Agreement 2017, 202-203).
Two-way contracts are not included in the team’s payroll. This means two-way player salaries are
excluded from team’s salary and thus, teams are not required to have any room in their salary caps
or exceptions to sign, acquire or convert a player to a two-way contract (Collective Bargaining
Agreement 2017, 192-193). Disabled player exceptions allow teams to sign or acquire one player
to replace a player, who as a result of disabling injury or illness is unable to render playing services.
(Collective Bargaining Agreement 2017, 200-202). Veteran free agent exception or Larry Bird
exception as commonly referred to consists of two, qualifying veteran free agent and non-
qualifying veteran free agent. The qualifying veteran free agent exception or Bird exception
enables teams to go over the salary cap in order to re-sign their own free agents up to the player’s
maximum salary. To meet the requirements for this exception player must-have played for three
seasons without clearing waivers or changing teams as a free agent. The non-qualifying veteran
free agent exception or Non-Bird exception allows teams to re-sign their players for the amount of
120% of the players last contracts final salary cap year plus 120% of any bonuses whether likely
or unlikely (Collective Bargaining Agreement 2017, 198-199). Minimum player salary exception
allows teams to sign a player or acquire by assignment a player contract not to exceed two seasons
in length. The salary provided for the player shall be equal to a minimum player salary that is
applicable to that player with no bonuses of any kind (Collective Bargaining Agreement 2017,

 12

207-208). With traded player exception, teams may for one year, following the date of the trade of
a player contract to another team, replace the traded player with one or more players (Collective
Bargaining Agreement 2017, 208-213).

These exceptions hold more rules to them, such as how much teams can pay for the players. All of
these salary cap exceptions show that it is possible to go over the given salary cap and be able to
pay larger sums to players.

1.2.3. Luxury tax

Luxury tax is a way to keep leagues is balance with respect to competitiveness. Collective
Bargaining Agreement (CBA) predetermines a surcharge that teams are obligated to pay if they
are to exceed a certain limit over salary cap. The reason why luxury tax exists follows the idea that
the player salary growth would slow down as well as the fact that it would prevent larger franchises
from acquiring the best players in the league (Dietl et al. 2009). NBA implemented the luxury tax
to the league in 1999. It is a tax system that forces teams with an average team payroll that exceeds
the salary cap by a predefined amount, to pay 100% tax to the NBA for each dollar the team’s
payroll has exceeded the tax level. The money that NBA obtains from teams who are obligated to
pay luxury tax goes to franchises that are financially weaker (Dietl et al. 2009).

1.3. Superstar effect

NBA is often labeled as a superstar-driven league. This is mostly due to the fame and skills of the
league’s top players which include players such as LeBron James and Stephen Curry (Gupte et al.
2019). These stars can bring enormous benefits to a franchise other than just performing extremely
well on the basketball court (Jane 2016). The Golden State Warriors won their first championship
since 1975 in the 2014-15 season and have won two additional championships since that, in the
2015-16 and 2017-18 season. Golden State Warriors gate receipts have increased from $55 million
in the 2013-14 season to $164 million in the 2017-18 season (Statista 2019). For NBA teams to
win championships they have to have to right players and the right team. In the current season
2018-19, the Golden State Warriors have five players from the NBA All-Star team. NBA All-Star
team consists of players that have been chosen to the team by fans, players, media and head
coaches. Usually the players that get picked to the NBA All-Star team have been performing
outstandingly during the season but also most of the chosen players are themselves so called
“superstars”. The Golden State Warriors having five NBA All-Star players attracts a huge number

 13

of fans all over United States as well as tourists from around the world. This has created the
massive increase of gate receipts for the Golden State Warriors over the past five seasons.

The superstar effect is created by an individual being able to attain considerable prominence and
success in their field. The earnings of a superstar are much higher than the earnings of their
competitors (Adler 2014). Sherwin Rosen is considered to define the phenomenon of stars by
saying that superstars are a relatively small amount of people who make more money and dominate
the field that they work in. Rosen explains the superstar phenomenon as follows. A person that is
slightly more talented than their competitors will make a greater earning that the competitors who
are slightly less talented. Why have person who is less talented than another, the less talented
person is simply put a poor substitute for a greater talent (Adler 1985). Moshe Adler argues against
Rosen’s explanation that superstar effect occurs where consumption requires knowledge. Mosher
argues: “As an example, consider listening to music. Appreciation increases with knowledge. But
how does one know about music? By listening to it, and by discussing it with other persons who
know about it. In this learning process lies the key to the phenomenon of stars” (Adler, M. 1985,
Stardom and Talent).

NBA superstars are defined by their actions on and off the basketball court. Not all great NBA
talents are considered as superstars as well as not all superstars are the greatest NBA talents. The
impact an NBA player can have off the basketball court can make them have the superstar effect.
At the end everyone has their own definitions of what NBA superstar is. There are some certain
factors that clearly show that someone has that superstar effect. These can be seen in team ticket
sales as well as ticket prices, the number of followers that player has on social media, actions on
the court and others.

1.4. Player efficiency

Basketball is a team sport where team performance is measured mostly by teams wins and losses
during a season. For a team to be performing well and getting as many wins in a season as possible,
they have to have the right players. Each individual player on a team gives something to the team,
some more and some less. The problem is that, it is difficult to measure how well an individual
player is performing and how efficient they are on the basketball court since basketball is not a
single player game (Van Curen 2012). Basketball on the other hand is one of the sports were an
individual player can have a huge impact on the end result of the game. When measuring player

 14

efficiency there are mathematical calculations that can be used in order to get a number which tells
how efficient an individual player has been for a team that they play in. The most well-known
ratings for player efficiency are, John Hollinger’s, Player Efficiency Rating (PER), Spanish ACB
League’s, Performance Index Rating (PIR), and NBA’s Player Impact Estimate (PIE).

John Hollinger’s PER is probably the most well-known and used rating from the three earlier
mentioned. PER measures player’s per-minute productivity and is pace-adjusted. The fact that the
rating is per-minute measure allows players to be compared even if they have not played the same
number of minutes during a season. Also, pace-adjustment makes it possible for each player to
have a fair PER score even if their team is more slow-paced then other teams. PER takes players
individual accomplishments such as field goals, free throws, 3-pointers, assists, rebound, block
and steals, as well as negative ones such as missed shots (field goals, free throws, 3-pointers),
turnovers and personal fouls (Hollinger 2011). PIR metric is used in the Euroleague and the
Eurocup as well as in many European basketball leagues. PIR metric has been criticized because
of the fact that it does not take into account aby weighting system that would determine the
importance of individual stats unlike PER and PIE rating (Wikipedia 2019). PIE is the most recent
player efficiency rating from NBA and is being used on NBA’s player stats. PIE is a simple metric
that gives a performance indication at both team and player level. The former rating that NBA used
was called efficiency (EFF), PIE is a major improvement to the former rating. PIE includes
personal fouls to the rating and NBA has added a denominator to the PIE as well. According to
NBA, the denominator removes the need to consider pace in the rating since the denominator acts
as an automatic equalizer. PIE measures the overall statistical contribution of a player against the
total statistics in games they play in. The result that PIE gives is comparable to other advanced
statistics such as PER, using a simple formula. PIE calculation takes into account individual player
points, field goals, free throws, rebounds, assists, steals and blocks, and also negatives such as
missed shots, personal fouls and turnovers. PIE also takes into account all the above-mentioned
stats but from the overall game (NBA Stats 2019).

 15

2. METHODOLOGY

This thesis follows Stanek (2016) paper in which Stanek explores player performance and team
revenues in order to come up with an NBA player salary analysis. This research explores the
relationship between player salary and player performance in the National Basketball Association
(NBA).

This research takes into account player performance during the regular season as well as the
playoffs. It is important to include both of these since NBA regular season starts in October and
ends in April having 82 games. The season has so many games that players are not necessarily able
to perform at their best during every game. Adding the fact that only 8 teams from each conference
gets a playoff spot. This is 16 teams out of the 30 teams total. The teams that know they have no
shot in making the playoffs might not include their top players in the last games of the season but
instead give more minutes to their bench player’s. Also, teams that are on the bottom of the league
might start to tank, which simply means teams start losing games in order to get the first pick in
the upcoming NBA draft. In the NBA draft the worst team of the season has the highest chance of
landing the number one pick. In the playoffs instead when the championship is on the line, players
tend to perform at a much higher level. The teams that get into playoffs have only one thing on
their minds and that is to win the championship. During the playoffs, better players tend to get
more time on the court and might even play the whole game of 48 minutes.

The comparison between regular season performance and playoff performance is truly necessary.
It does not always mean that when the playoffs start that each and every player increases their
performance. For example, Damian Lillard who plays for the Portland Trailblazers had a Player
Impact Estimate (PIE) of 16,5 during the regular season. During the playoffs Lillard’s PIE was
only 6,2 and their team lost the first four games in the first round of playoffs ending their season.
This does not mean that Lillard was the only reason for the losses but to show that even though
playoffs is the time when players tend to increase their performance it is not always the case. Other
player’s like LeBron James had a PIE of 19,1 during the regular season and in the playoffs his PIE

 16

increased to 22,4. James led his team to the NBA finals where they lost the series to Golden State
Warriors.

The data used in this research is compiled from Spotrac as well as NBA’s own stat website. Spotrac
is a website founded by Michael Ginnitti and Scott Allen which provides team and player contracts,
salaries, salary cap management, payrolls as well as other information for North American leagues
NFL, NBA, MLB, NHL and MLS. Spotrac has the latest data about NBA team payrolls and player
salaries that are used in this research. NBA stats on the other hand is a website that provides
information about performance statistics of NBA teams and players. NBA stats has the latest data
of team and player performance that are used in this research. The data used in this research paper
observes 361 player’s that played in the 2017-18 season. The players were limited so that a player
had to play at least 30 or more games during the season as well as average 7 minutes of play per
game. The data was limited to this quantity in order to get more accurate answer about the
relationship between player salary and player performance. Since player’s might get injured before
or during the season it is not fair to examine their relationship between salary and performance.
Also, the playoff performance data was limited so that players had to play at least in four games
during playoffs since four games is the minimum amount of games a team can play in the playoffs.

This research observes the season 2017-18 and the team payrolls and player salaries are collected
from Spotrac which relies on collecting their data about player salaries from Basketball Insiders.
Even though Spotrac has become one of the largest online resources for team and player contracts
it should be remembered that the data from Spotrac is collected from another source and therefore
has to be treated in the right way as it may have some errors. The data for team and player
performance that comes from NBA stats and is collected by NBA itself since they keep track of
the stats that happen during games, so this source can be considered as reliable. The data from
NBA has various specific statistics about player shooting, assists, rebounds, turnovers etc. The
most important statistic for this paper is the PIE measurement which measures player player’s
overall statistical contribution against the total statistics in games (NBA stats, 2019). PIE formula
is as stated below (NBA stats, 2019). See definitions in appendix table 1.

 = ( + + − + + (. 5 × ) + + + (. 5 × ) −
 − )/( + + − − + + (. 5 ×
 ) + + + (. 5 × ) − − ) (3)

 17

This paper measures player performance with Performance Impact Estimate (PIE) during the
regular season as well as playoffs. This measurement gives each player a number that shows their
PIE score. The aim of this research is to see how player salary contributes to player’s PIE score. It
has to be remembered that player’s salary varies with age and experience. Players who have just
entered the league usually have a rookie contract which is much lower than player’s that have 10
years of experience. Age and experience do not only determine player salary, but usually more
experience leads to a higher PIE score. Again, it is not the case that older more experienced player’s
necessarily have higher PIE score’s but in the usual sense performance tends to increase with
experience. In addition to taking into consideration age this research takes player position and team
in to account. Table 1. shows descriptive statistics gathered from NBA stats and Spotrac.
Performance metrics PIE and Playoff PIE display the differences that NBA players have in
performance. These two metrics have been taken from NBA stats. Cap percentage (Cap %) which
is the percentage of players salary contribution to the team payroll and base salary (Base Salary)
which is the actual player salary in millions USD are taken from Spotrac. Age shows the age range
in the NBA and it is taken from NBA stats.

Table 1. Descriptive variable statistics
 PIE Playoff PIE Cap % Base Salary Age

 Observations 361 166 361 361 361
 Mean 9,67 8,83 0,068 7,62 26,2
 Std. Dev 3,17 4,84 0,067 7,63 4,3
 Min 1,7 -7,3 0,0001 0,01 19
 Max 19,4 26,6 0,289 34,68 40

Source: Author’s calculations

Previous studies have shown that the determinants for player salary seem to be points, rebounds,
personal fouls and experience (Compton, W., Sigler, K. 2018). This research uses OLS regression
where PIE and Playoff PIE are used as the dependent variables. Independent variables consist of
cap percentage, base salary, age. The regression also applies fixed effects, position and team to
estimate the effect of salary to player performance.

To study furthermore the relationship between player salary and performance each team’s total
PIE is being calculated by adding up the PIE scores from each player in a same team. From there
each player’s PIE score is divided with total team PIE in order to find out what has been the

 18

contribution of each player to the team PIE. This gives a result that can be compared to each
player’s actual cap percentage, the percentage amount that a player earns from their team’s total
payroll. Same calculations will be added to playoffs. The result from comparing the player’s PIE
contribution to their salary cap percentage will give suggestions about how much players are
possibly under- or overpaid in relation to their performance on the court. The first assumptions are
that younger less experienced players are underpaid in relation to their performance and older more
experienced are overpaid. The equation for this calculation is stated as:

 RSTUVW RXY
 = % − ZVT[ RXY
 (4)

 19

3. EMPIRICAL PART

This research aims to explain the relationship between player salary and player performance. It is
important to understand that the results given in this research do not determine whether salary
always leads to a better performance. Since player performance is calculated on an individual level
this research does not take into consideration the fact that the presence of a certain player on the
court can improve other’s performance. This kind of presence on the court also contributes to the
salary that a player receives. To examine the relationship between salary and performance this
paper incorporates age and fixed effects position and team into the regression to gain a better
understanding on how salary actually affects performance.

3.1. Regular season regression

Regular season performance might vary since each team plays 82 games during the season. It is
still important to look at the relationship between player salary and player performance during the
regular season because ultimately the regular season decides which teams get to play in the
playoffs. Most players can keep a quite solid season performing well in each game. Some players
might have a good start of the season and then performance starts to decline and vice versa.

The first regression takes player’s share of salary, the percentage amount that a player earns from
the team’s total payroll, into consideration. The share of salary in the regression is named CAP.
The NBA has a salary cap but due to different exceptions mentioned earlier in part 1.2.2. teams
are allowed to go over the salary cap. This means that teams can have different sized payrolls and
so, for example cap percentage of 20% can be different amount in different teams. In model 1 the
regression takes into account player’s age in addition to the share of salary. The second model
takes the fixed effects position and team into consideration. The sample size for this regression is
361 observations.

 20

Table 2. Share of salary and PIE regression
 Model 1 Model 2
 Variable Coef St. error Coef St. error
 const 9.87*** (0,92) 9.00*** (1,25)
 CAP 22.94*** (2,33) 21.80*** (2,26)
 AGE -0.07* (0,04) -0.09** (0,04)
 Position FE NO YES
 Team FE NO YES
 N 361 361
 Adj. R2 21 % 30 %
 Note: *** - p

percentage. Model 4 applies fixed effects position and team into the regression again. Base Salary’s
relationship with PIE remains quite similar to model 3 and Age remains insignificant. Adjusted R
squared explains 22% in model 1 and 31% in model 2.

3.2. Playoff regression

This part of the research examines the relationship between player salary and playoff performance.
Players tend to perform better during the playoffs and so, there should be a more positive
relationship between salary and playoff performance than what salary and regular season
performance has. Playoff regression has 166 observations which is almost 200 observations less
than the regular season. These observations include the player’s that did make it to the playoffs.

Table 4. Share of salary & Playoff PIE regression
 Model 5 Model 6
 Coef St. Error Coef St. Error
 const 7.30*** (2.21) 8.25*** (3.01)
 CAP 26.70*** (4.94) 26.26*** (5.07)
 AGE -0.021 (0.08) 0.002 (0.09)
 Position FE NO YES
 Team FE NO YES
 N 166 166
 Adj. R2 15 % 15 %
 Note: *** - p

Table 5. Base Salary & Playoff PIE regression
 Model 7 Model 8
 Coef St. Error Coef St. Error
 const 7.53*** (2.21) 8.67*** (2.99)
 BASESALARY 0.23*** (0.04) 0.23*** (0.04)
 AGE -0.03 (0.08) -0.002 (0.09)
 Position FE NO YES
 Team FE NO YES
 N 166 166
 Adj. R2 15 % 16 %
 Note: *** - p

Table 6. Suggested cap percentage

 Age Suggested Cap %

 Mean -1.47%

 Mean ≥ 24 -4.64%

 Mean 25-29 0.70%

 Mean ≤ 30 0.59%
Source: Author’s calculations

Table 6. shows the suggested cap percentage that players should have according to this method.
Player’s in the NBA on average are underpaid 1.47% according to this model. The most underpaid
player’s in relation to their performance are found to be players who are 24-year-old or under who
are on average underpaid by 4.64%. This is most probably because of rookie contracts which are
much lower in value than veteran contracts. Player between 25-29 years old are on average are
overpaid by 0.70% and player’s over 30 years old are on average overpaid by 0.59%. What these
results suggest is that on average each player in the NBA should have cap percentage increased by
1.47%.

3.4. Limitations

The results showed that there is a positive relationship between player’s salary and. Still it is
important to look at some of the limitations that this research has. Firstly, this research only takes
individual performance into consideration and not the team as a whole. Even though player’s PIE
score might be low, it does not mean that they are not playing well or contributing to the team.
Many teams employ veteran players who have many years of experience in the NBA. Why teams
do this is because, they have been in the league for many years and they can help young player’s
entering the league on and off the court. In these cases, usually the veteran salary is quite low as
well, so their salary and performance go pretty much hand in hand. Also, some player’s might have
a great off-ball game, movement on the floor without the ball in their hand, which does not
contribute to their PIE score but does to the team’s overall game.

Superstar effect is another thing that this research does not take into consideration. Superstar effect
does not change the fact that salary and performance have a positive relationship but when it comes
to the suggested cap, players bring much more to their teams than just what they do on the court.

 24

Many big-name players who are considered to be superstars might be somewhat responsible for
the tickets sold to a game. These superstar player’s might perform not according to their salary,
but they contribute to their salary in other ways. For example, LeBron James played for the
Cleveland Cavaliers in 2017-18 season. Cleveland Cavaliers during that season had a total
attendance of 843,042 people in 41 games and their attendance was the second highest in the whole
league. Next season LeBron James changed his team to Los Angeles Lakers and Cleveland
Cavaliers attendance for the season 2018-19 dropped to 793,337 people in 41 games (ESPN,
2019). This shows that one superstar player can attract a lot of people to come and watch games
and at the same time generate ticket revenues for their team. Other value that superstar players
might bring to a team are revenue from sold jerseys and sponsorship.

Limitation in this research is also the fact that only one season is being observed. This research
only examines the relationship between player salary and performance from one season, which
means it does not show if in the past the relationship has been different. Having more season would
lead to results that would show if the player salary and performance has had a positive or negative
relationship throughout the seasons. Although only one season is being examined, it should give
rather accurate results from the relationship.

Lastly what could be considered as a limitation is that the player’s in the NBA earn higher salaries
as they get more experience. That is why the suggested salary cap percentage shows that players
who are 24 years old or younger according to the study are underpaid. This is due to the fact that
rookie contracts are much smaller in comparison to what veteran contracts are. As young players
enter the league they might be performing at a high level while having a low salary. Older and
more experienced player’s might have large salaries and their performance starts to decline as they
get older. In addition to this, the author’s model shows the average over- or underpayment of
players.

 25

CONCLUSION

The salary of an NBA player has a positive relationship to their on-court performance. However,
is every NBA player paid according to their performance? Professional sports around the world
are huge business and NBA is becoming the biggest sport in North America. NBA franchises are
generating more revenue each year which increases the salary cap for teams. This increase in salary
cap allows teams to offer bigger contracts to players. Still, is it the case that these big contracts,
from which NBA players earn whopping salaries have a positive relationship to their on-court
performance. Furthermore, do the players in the NBA actually perform according their given salary
or should they be paid even more. This research examines if there is a positive relationship with
player salary and performance as well as if players are paid according to their performance.

The regressions done in this research shows that there is a significant relationship between player
salary and performance. This means that more player earns, the better they will perform on-court.
Regression also examined the relationship with salary and playoff performance, since playoff
performance usually tends to be better than regular season performance. The results showed a
significant relationship as well. Players salary relationship to Playoff performance was not that
much more positive than regular season performance which did not come as a surprise.

Author’s own model examined if players in the NBA are paid according to their performance on-
court. The results obtained from this model showed that on average players are underpaid, however
this underpayment of players mostly comes from young players. Players who are 24-years old or
younger are performing better in relation to their given salary. This can be explained by the rookie
contracts in NBA which limits the amount of salary these young players are able to earn. As players
get older, they become overpaid according to the model. This model may indicate that teams should
allocate their payroll more evenly amongst their players. On the contrary, this model does not take
into consideration superstar effect which in most cases is a reason to pay more for certain players.

This research shows a significant relationship between player salary and performance as well as a
possible underpayment of players according to author’s model, however, there are certain

 26

limitations to notice. The performance metric PIE, which is used in the regressions and author’s
own model only accounts for individual statistics. PIE score does not bring justice to every player
since player’s PIE score might be low, but they are an important force on-court otherwise.
Superstar effect is another limitation this research has. Players with superstar effect usually are
performing great on-court but, in addition to their great on-court performance they are a major
asset for teams. Superstars bring large amounts of revenue for their teams only with their presence.
This is something that teams take into consideration in the contracts they offer for these superstar
players. Lastly, this research only observes one season. The general idea is that higher salary has
a positive relationship between performance and this research showed that there is. However,
observing only one season does not tell if this has always been the case.

This paper explored the relationship between player salary and performance as well as the possible
over- or underpayment of players. Player salary does not only consist of the performance they
bring on-court. Therefore, this suggests that in future researches not only players performance
should be taken into consideration but everything else they bring to a team. By including superstar
effect which can indicate how much visibility and revenue a player brings to a team can give a
better understanding on how player salary is determined.

 27

LIST OF REFERENCES

Adler, M. (1985). Stardom and Talent: The American Economic Review, Vol 75, No. 1, pp. 208-
 212. Accessible:
 https://www.jstor.org/stable/pdf/1812714.pdf?refreqid=excelsior%3Ad95105fbb02f36f
 60090a08636cea39c, 5 April 2019.

Adler, M. (2014). Stardom and Talent. Accessible:
 http://www.columbia.edu/~ma820/Stardom%20and%20Talent.pdf, 5 April 2019.

Caldwell. T. (2010). An Overview and Comparative Analysis of the Collective Bargaining
 Agreements in the NBA, NFL, and MLB. Accessible:
 https://scholarship.claremont.edu/cmc_theses/62/, 2 April 2019.

Collective Bargaining Agreement. (2017). Accessible:
 http://3c90sm37lsaecdwtr32v9qof.wpengine.netdna-cdn.com/wp-
 content/uploads/2016/02/2017-NBA-NBPA-Collective-Bargaining-Agreement.pdf, 2
 April 2019.

Compton, W., Sigler, K. (2018). NBA Players’ Pay and Performance: What Counts? Accessible:
 http://thesportjournal.org/article/nba-players-pay-and-performance-what-counts/, 12
 April 2019.

Coon, S. (2018). Larry Coon’s NBA Salary Cap FAQ: 2017 Collective Bargaining Agreement.
 Accessible: http://www.cbafaq.com/salarycap.htm, 5 April 2019.

Dietl, H. M., Lang, M. Werner, S. (2009). The Effect of Luxury Taxes on Competitive Balance,
 Club Profits, and Social Welfare in Sports Leagues. Accessible:
 http://www.isu.uzh.ch/static/ISU_WPS/91_ISU_full.pdf, 5 April 2019.

ESPN. (2019) NBA Attendance Report – 2019. Accessible:
 http://www.espn.com/nba/attendance, 25 April 2019.

Gupte, C., Kaplan, S., Premkumar, D., Ramamoorthy, V., Sagar, A., Wilbur, J., Zilberman, D.
 (2019). The Economic Impact of NBA Superstars: Evidence from Missed Games using
 Ticket Microdata from a Secondary Marketplace. Accessible:
 http://www.sloansportsconference.com/wp-content/uploads/2019/02/Economic-Impact-
 of-NBA-Superstars.pdf, 5 April 2019.

Hollinger, J. (2011). What is per? Entertainment and Sports Programming Network. Accessible:
 http://www.espn.com/nba/columns/story?columnist=hollinger_john&id=2850240, 5
 April 2019.
Huang, J. (2016). Salary in the National Basketball Association. Accessible:
 https://repository.upenn.edu/cgi/viewcontent.cgi?referer=https://www.google.com/&htt
 psredir=1&article=1009&context=joseph_wharton_scholars, 12 April 2019.

 28

Humphreys, B. R., Johnson, C. (2017). The Effect of Superstar Players on Game Attendance:
 Evidence from the NBA. Accessible: https://business.wvu.edu/files/d/f79a6799-edcc-
 4962-9b9f-57d8134f65d3/17-16.pdf, 5 April 2019.

Jane, W-J. (2016). The Effect of Star Quality on Attendance Demand: The Case of the National
 Basketball Association. Accessible:
 https://journals.sagepub.com/doi/pdf/10.1177/1527002514530405, 5 April 2019.

McCartney, H. (XXXX). Do Sports Stars Deserve Their Wages? Should Their Governing Bodies
 Introduce a Wage Cap? Accesible:
 https://www.uea.ac.uk/documents/953219/6293155/McCartney_UG1_First+Place.pdf/2
 1ac8f12-8270-486e-afbd-70bd77c935db, 12 April 2019.

NBA Official. (2017). 2017 NBA Collective Bargaining Agreement: Principal Deal Points.
 Accessible: https://official.nba.com/2017-nba-collective-bargaining-agreement-
 principal-deal-points/, 2 April 2019

NBA Stats. (2019). Advanced Stats. Accessible: https://stats.nba.com/, 12 April 2019.

NBA Stats. (2019). Frequently asked question. What is PIE? Accessible:
 https://stats.nba.com/help/faq/, 5 April 2019.

Sackley, W. H., Sigler, K. J. (2000). NBA Players: Are They Paid for Performance?, Managerial
 Finance, Vol. 26 Issue: 7, pp. 46-51. Accessible:
 https://doi.org/10.1108/03074350010766783, 12 April 2019.

Shorin, G. (2017). Team performance versus performance in professional sports: Is increased
 spending associated with greater performance? Accessible:
 https://dukespace.lib.duke.edu/dspace/bitstream/handle/10161/14332/Shori
 n2017.pdf?sequence=1, 2 April 2019.

Spotrac. (2019). NBA Salary Rankings: 2017 Base Rankings. Accessible:
 https://www.spotrac.com/nba/rankings/2017/base/, 12 April 2017.

Spotrac. (2019). NBA Team Salary Cap Tracker 2017-2018. Accessible:
 https://www.spotrac.com/nba/cap/2017/, 12 April 2019.

Stanek, T. (2016). Player Performance and Team Revenus: NBA Player Salary Analysis.
 Accessible:
 https://scholarship.claremont.edu/cgi/viewcontent.cgi?referer=https://www.google.com/
 &httpsredir=1&article=2302&context=cmc_theses, 12 April 2019.

Statista. (2019). Gate receipts of the Golden State Warriors (NBA) 2017/18. Accessible:
 https://www.statista.com/statistics/381541/gate-receipts-of-the-golden-state-warriors-
 national-basketball-association/, 5 April 2019.

Van Curen, A. (2012). The Effect Alternate Player Efficiency Rating Has on NBA Franchises
 Regarding Winning and Individual Value toa n Organization. Accessible:
 https://fisherpub.sjfc.edu/cgi/viewcontent.cgi?referer=https://www.google.com/&httpsre
 dir=1&article=1037&context=sport_undergrad, 5 April 2019.

 29